Adoption, not innovation, biggest gap in AI rollout: Experts at AI Summit

Experts at the India AI Impact Summit 2026 said the biggest hurdle in using AI for development is not technology but adoption, weak capacity and lack of investment in digital public infrastructure

Adoption challenges, weak institutional capacity and lack of investment in digital public infrastructure are the biggest barriers to scaling artificial intelligence (AI) for development, experts said at the India AI Impact Summit 2026.

 

Speaking at a session titled ‘AI for Impact — Global South Forum: Shared Solutions, Shared Futures’, industry leaders highlighted why promising AI strategies often fail to translate into real-world impact.

 

Janet Zhou, director at the Gates Foundation, said the primary gap today is not innovation or evidence, but adoption. "The biggest gap is adoption, although I don't know if I would frame it solely as a lack of capacity. I think for philanthropy, one of the goals is always to make adoption easier."

 

 

"There are two pathways for that. One pathway is building shared infrastructure so that markets can work better to serve vulnerable people. It involves streamlining market entry to make it easier for low cost suppliers to enter and compete," Zhou added.

 

According to Zhou, another key approach is using AI to directly tackle major social challenges. "The other pathway is to take AI and point it at the problems, so that if those problems were solved, adoption would feel inevitable," she said.

 

Why AI plans fail in execution

 

Shalini Unnikrishnan, managing director and senior partner at Boston Consulting Group, said most AI transformation failures stem from misplaced priorities.

 

She talked about the 10/20/70 principle in AI, which states that 10 per cent of the process lies in algorithms, 20 per cent in process changes, and 70 per cent in capacity and adoption.

 

According to Unnikrishnan, the funding patterns worsen the problem. "Companies extract much more time and energy talking about the 10 per cent, that they do not focus on the adoption. This becomes the fundamental disconnect between things looking good on paper and then never getting implemented and adopted at scale," she said.

 

Unnikrishnan stressed two priorities for governments. One, supporting the governments to articulate their own vision, versus it being a vision handed to them. Second, understanding that AI needs time and investment to go forward.   

 

Institutional capacity lagging behind ambition

 

Robert Opp, chief digital officer at the United Nations Development Programme (UNDP), said countries are eager to use AI but struggle to scale.

 

"Every country is interested in how to embrace AI for national development. AI represents one of the most powerful potential accelerators for achieving goals," Opp said.

 

"But the challenge there is that the ambition outstrips the capacity of institutions, and the overall setup that would be required to actually implement these things at scale," he added.

 

Opp emphasised the need to invest in foundational systems. "Investing in the shared building blocks, digital public infrastructure that will allow AI to scale at a population level is important. Getting data sources, data sets and data governance is important and has high leverage," he said. 

 

Pilot projects often fail to scale

 

Johan Harvard, global AI advisory lead at the Tony Blair Institute, said governments frequently underestimate deployment challenges. He said that resources should focus only on scalable projects.

 

"We need to help governments, and we need to work with donors to make sure to focus on resources that help in scaling," Harvard said.